any word on gnutella 2 adoption?

[CaptainAvitar]

This is really weird I really thought that the new protocols that Shareaza started to use would cause a big stir. Theres not alot of talk about it on the Limewire site, in fact I hear that the code will be released this week or next week sometime. Do you think other gnutella clients (LW,BS, etc) will acept it ? if so how long would it take to put into thier programs?

[jonny5]

interesting questions you have raised.. I assume if the gnutella 2 network is proven to run more efficiently and quicker, then the other clients will do what shareaza is doing and use both g1 and g2. I didn't know/haven't really been paying attention.. but you said that g2's source code hasn't been released yet.. so that's probably why there has been no talk of it on other gnutella based program's websites.

That's just what i think.. my ramblings at 1am :P
I'm interested in seeing what other people think too.
*thinks this is a good thread* :)

[evilmegaman]

The only thing is that even when I am just connected to gnutella 2, I get uploads from many different clients from the gnutella 1 network so I imagine that gnutella 2 is only a portion of the network.

[Sephiroth]

There has been discussion on the GDF (http://groups.yahoo.com/group/the_gdf/) about it..

One of the best posts about it is Greg Bildson's post (http://groups.yahoo.com/group/the_gdf/message/11608) which i think probably best summarizes all the other gnutella developers opinion on it.

Personally i think shareaza damaged their relationship with how they have handled "gnutella2" so far. They didnt tell any developers before which all gnutella proposals that have been agreed and are in gnutella now have gone through the GDF and havent been done in the ship it now, and get other programs to jump on the bandwagon.

IMHO Mike has given up on the GDF and assumed that they wouldnt listen to his idea or have allready rejected his idea which is why Shareaza and mike has said that he will support other implementations also. I think those kind of assumptions are wrong.

Also hes hasnt handled this entire thing in the best of manners in my opinion and the fact that they are trying to capitalize off the name "gnutella" by using "gnutella2" without taking the traditional route in proposing protocol changes and without the support of the gnutella community behind them was wrong. Its "Gnutella2" without "gnutella." Weather its Mike or someone thats helping him out its mike's program, hes in charge and weather he likes it or not it all comes back to him..

Also by pushing people from the shareaza forums to go out on the gdf and elsewheres to "spread the word" which is only hurting shareaza. Blowing this up into a big flame war wont accomplish anything.

The idea that gnutella just needs "marketing" is dumb and really nothing more would please me than to have more people using gnutella but the approach that shareaza has taken IMHO was not the best way at all. Because they have fragmented themselves from gnutella on what should be a common goal. Find me a developer that would not want a better way to search on gnutella that is more efficent than broadcasting and i suspect youll have a very hard time..

The other devs arent stupid they arent going to commit to something that they havent seen. Yeah shareaza may or may not release the specs but i think its wrong for people to expect the other devs to make a decision that they will support it or not until then especially since none have committed to it.

No matter how well it works in beta becuase the number of users is too small to see how well it work if a very large number of people used it. People dont understand that if gnutella only had 70 thousand users or so then it would work just as great as "g2" or whatever in that you could search every single host and it would be alot quicker. People assuming that it will work and is flawless is pretty premature too.

Me im waiting until what the people who have been working on gnutella for years have to say until i determine how good or bad it is. If it doesnt get accepted then i dont think that people should blame the other developers because they are really looking out for the best interest in gnutella and not just letting "marketing" and "the mob" affect their judgement.

Ive probably pissed some people off here but i just want to realize that Im not against shareaza or anything. I just disagree with what most things that shareaza is doing which they think is helping gnutella.. I might clean this up and maybe wait a little bit to see what happens and post it all as a editoral too.

[John W. Lindh]

Originally posted by CaptainAvitar
Do you think other gnutella clients (LW,BS, etc) will acept it ?
After Shareaza Mike pissed them off? No. - I would even have doubted that they would've implemented it completely, if it had been proposed in an acceptable manner - but they might have implemented one part or another.

[Rickio]

I think it will all add up to the facts. It will either work will and others will follow, or it will fail and be ignored.

Bearshare tried a little move awhile back, didn't seem to make them all that better.

Might not have been to smart to make a move (to step around gnutella development) as drasitc as they did, but let the chips fall where they may now.

[Borgster]

im beginning to think that if other clients like bearshare, and limewire dont connect to the gnutella 1.0 (G2) protocol, that shareaza may use that as there own network, and shareaza can have there own personal network to run clients off of, like napster did.

[Sephiroth]

Originally posted by Borgster
im beginning to think that if other clients like bearshare, and limewire dont connect to the gnutella 1.0 (G2) protocol, that shareaza may use that as there own network, and shareaza can have there own personal network to run clients off of, like napster did.

Shareaza is allready on its own propierty network because it has not been supported by the other developers mainly because they havent seen it and just because a program uses the gnutella name doesnt make it gnutella.

Where did you get gnutella 1.0 from? Im just wondering.

[Nothingface5384]

i heard the phex will be supporting G2 and hopefully when blues army tcourse s over..bout 4 weeks left he said..he'll work on his peerahna and his network some more..hopefully adding g2 to peerahna also!

[Crazy Horse]

Good post Seph. I can't pretend to think that I understand all the technical stuff that goes into making the Gnutella network. Nor do I understand what all the developers are doing. I do know that Gnutella needs to scale to meet the demand of a larger user base. I do understand Greg concerns in that matter. The jury is still out on whether this G2 thing is going to be beneficial and only time will tell how good this truly will be. The cat's out of the bag though and like Rickio said let the chips fall. A lesson will be learned - good or bad. I'm hoping this is a major step in the right direction for Gnutella. So many developers should be coming up with major innovations sooner than later. Maybe Mike just upped the ante.

[Rickio]

Originally posted by Nothingface5384
i heard the phex will be supporting G2 and hopefully when blues army tcourse s over..bout 4 weeks left he said..he'll work on his peerahna and his network some more..hopefully adding g2 to peerahna also!

I tried phex and peerahna in the past and they seemed decent. I think it's definately a boost in the new direction if more clients implement the new features.
I can only feel it is positive, but time tell....

peace

[evilmegaman]

One thing is that gnutella2 is like shareaza's way of doing what bearshare does. I think it is just a portion of the gnutella network as I still get uploads from clients like gnucleus 1.6.0.0 and others so there is something very sneaky about gnutella2. I think that mike isn't like he used to be at all... He's trying to hog the glory... he has become a bit overwhelmed... It is very sad for me to say this but... I think he is lying about gnutella2.... I think it is shareaza's portion of gnutella like bearshare did. you can still upload to other clients but you can only connect to a small number of other things...

in reply to nothingface, He is a little skeptical (http://www.gnutellaforums.com/showthread.php?s=&threadid=17017)

[Nothingface5384]

mega man - i think its just a seperate protocol thats backwards compatiable w/ the current gnutella network....

[crackerjacker]

heh propietary network, how can it be propietry if no one owns the gnet *
hmmm
heh but one thing comes to mind,
note* i aint flaming no one on this thread*
i just think that what seph said is because he felt to express what he saw gnutela as, and what it becomes.

can u prove shareaza is propietry.
no what a questeion, why would i ask it
in part excerpt from this*
whatever in that you could search every single host and it would be alot quicker.

who can attest to this
if this is the case, searching every single host using the latest version of shareaaza*
this will be fuking amazing.
more hosts and sources files for many*\

later

[Borgster]

Originally posted by Sephiroth
Where did you get gnutella 1.0 from? Im just wondering.

you know on how the gnutella clients connect to gnutella 0.4 and 0.6? 1.0 is basically another version of the protocol after 0.6.

[John W. Lindh]

Originally posted by Borgster


you know on how the gnutella clients connect to gnutella 0.4 and 0.6? 1.0 is basically another version of the protocol after 0.6.

The latest gnutella version is 0.4. Gnutella 0.6 is only a draft at rfc-gnutella.sf.net yet. - Gnutella2 is NOT Gnutella 1.0. It is not compatible to Gnutella and so at the moment it's just a protocol by a client that also implements Gnutella.

[Anenga2]

Here is some FAQ for you.

Why did Mike choose to release G2 as he did?
First of all, Mike has replied in the GDF on why he choose to release Shareaza with G2 as he did. He has good, legitimate, valid points. You can view his post here: http://groups.yahoo.com/group/the_gdf/message/11571

Where are the specs for G2?
Mike plans on releasing the G2 Protocol Spec to the GDF after v1.7 of Shareaza is complete (goes out of beta). I'm positive Mike will get the spec out in time.

The spec will be a proposed standard for Gnutella. G2, defined, is a new set of open standards. Yes, it is not open at the moment (because the specs haven't been released yet) but it will be soon. ETA for the specs is in around 1 week.

The reason Mike did not release the spec yet is becuase he wanted to test his G2 network in the wild with a large beta group. He's tested it, and everything looks solid.

Will other Gnutella clients adopt G2?
Regarding other clients, well... it's their decision if they want to use G2 or not.

If you want to know if your favorite Gnutella client will be supporting G2 or not, ask them. They probably won't have an answer for you until the specs are released. Even then, they might have issues with it. I'm positive the specs will hold up to the GDF's expectations. Of course, expect some changes to occur to G2 in the future.

Many developers have said they'll support it if is stable. Vinnie of BearShare said he'll support it if Limewire does, Limewire says it will support it if it holds up to their expectations, Gnucleus says they are eager to see the spec and to discuss it with the GDF. GTK's (the open source Linux client) developer has been active in the Shareaza IRC channel. So it seems like he will be interested in the spec when it is released as well.

Sure it works well now, but how about when you add 100,000 more nodes to it? Surely it can't hold up!
Seems like Mike is confident that it will scale even with that many nodes. Here is what he said in the Shareaza Forum in reply to one skeptical developer:


This is an interesting example. Lets say that there are 500,000 nodes on the network, sensibly organised into hubs and leaves. Shareaza hubs aim for 100-200 leaves, but for the sake of making it harder, lets say they have about 100 each. Thats about 4,950 hubs and the rest (495,050) leaves.

To begin with, every node on the network does a search an hour, which is pretty conservative, but its a start. Gnutella2 queries vary in length substantially, but 64 bytes is not a bad size I guess. 500,000 queries per hour is about 140 queries per second. At 64 bytes each thats very approximately ~8888 bytes. If every hub has to receive every query, every hub must have at least 70 kilobit per second inbound bandwidth. That is not a lot of bandwidth for a hub to have, in fact you have to have 144 kbit/s minimum before you can even become a hub.

And it actually gets better, because every hub does not have to receive every query to do a global search. And where a hub does process a query, the leaf filtering process is good enough that very few of the attached leaves will actually see it. Thus the passive bandwidth burden on leaves is much smaller than that experienced in Gnutella1. Active leaf bandwidth requirements (when searching) scale perfectly to however much bandwidth the leaf has available to it -- if you have less, it simply takes longer.

In conclusion, Gnutella2 easily supports true global searching on a 500,000 node network assuming one query per hour. As stated above, one per hour is very conservative. All optimisations aside, 2 per hour would require 140 kbit/s hubs. 4 per hour would require 280 kbit/s hubs, etc. At first that doesn't sound good, but there is good news. Not every hub has to receive every query, and hub by hub control means that it is possible to throttle per hub query traffic on the client side (eg: actively restricting per client per hub query traffic based on the capacity of the hub).

LimeWire's network size meter suggests there are currently about 100,000 nodes on Gnutella1. I don't know how accurate that count is, but assuming its not way off, the entire Gnutella1 network could move to Gnutella2 right now and be able to find any file anywhere.

How does G2 work with G1 in Shareaza?
G2 nodes cluster in Shareaza with a default value of 70%. That means 30% is G1 (Old Gnutella). If your a leaf, your connected to 1-3 G2 Hubs 3+ G1 Ultrapeers. So in actuality, your searching both Gnutella1 and Gnutella2. So you'll have the same results as you get in Limewire, Gnucleus etc. plus Gnutella2 (which is currently all of Shareaza).

So basically, all nodes on Shareaza support G1 and G2. Leafs search G1 and G2, Hubs have leafs which are G1 and G2 and have peers which are G1 and G2. Shareaza is not ditching G1.

If you have anymore questions, e-mail me at [email protected] or post something up in the Shareaza forums (www.shareaza.com/forums)

[Anenga2]

Originally posted by evilmegaman
One thing is that gnutella2 is like shareaza's way of doing what bearshare does. I think it is just a portion of the gnutella network as I still get uploads from clients like gnucleus 1.6.0.0 and others so there is something very sneaky about gnutella2. I think that mike isn't like he used to be at all... He's trying to hog the glory... he has become a bit overwhelmed... It is very sad for me to say this but... I think he is lying about gnutella2.... I think it is shareaza's portion of gnutella like bearshare did. you can still upload to other clients but you can only connect to a small number of other things...

in reply to nothingface, He is a little skeptical (http://www.gnutellaforums.com/showthread.php?s=&threadid=17017)
What are you talking about? Mike cares a lot about his users, that's why he spent a lot of time working on Gntuella2. It works very well, and will be proposed to the GDF soon enough.

Mike is not trying to hog anything. It's just his way of doing things, think, code, release to public, then release to the GDF. Developers may have problems with this, but Users/Fans shouldn't.

How is mike lying about G2? Care to elaborate?

[Sephiroth]

Originally posted by John W. Lindh


The latest gnutella version is 0.4. Gnutella 0.6 is only a draft at rfc-gnutella.sf.net yet. - Gnutella2 is NOT Gnutella 1.0. It is not compatible to Gnutella and so at the moment it's just a protocol by a client that also implements Gnutella.

The latest verison is Gnutella 0.6.

Anenga2 if your going to "clear things up" then you have to do more than just repeat what has allready been said but in a different form.

Also i think the last thing you should do is to speak for other developers who have not said that they will support it, it would be another stupid move by you to start telling people "oh theyll support it if its stable" and like i said it would be incredibly stupid for them to say that they will support anything without even seeing it..

Mikes post your referring too has a few problems. First limewire's host count is undercounting the network by about at least 4 times. So the real number of hosts on gnutella is 400000 users and its most likely much higher than that not 100000. So if gnutella moved to "gnutella2" then a person would not be able to search the entire network. It may support global searching on a 500000 user network but what about millions? After 500000 users it seems that "gnutella2" suffers many problems that the current network. How can gnutella compete with a network like fasttrack if they can support millions of users yet "g2" only guarentees up to support only 500000 and the current gnutella size is around that? So is it really that much better as you are marketing as it being? Keep in mind im baseing all this off the information in mikes post..

[kiwi_uk]

FT doesnt search the whole network. Searching 500000 nodes is more than most, if not all other networks can manage.

[Sephiroth]

Its not really known for sure since fasttrack limits search results. Even though fasttrack was probably a poor example I still would like to know given in a very large network how gnutella2 would be any different than gnutella today.

In any case one of the main parts of "gnutella2" is this marketing.. Which Gnutella does not need to be "rebranded" repackaged, sugarcoated and forced down people's throats, people dont like gnutella because they think its centralized and dont know anything about it. The majority of people who run around and say "gnutella sucks" have either never used a gnutella program ever or have not in a very very long time.

[rainwater]

Originally posted by Sephiroth
Its not really known for sure since fasttrack limits search results. Even though fasttrack was probably a poor example I still would like to know given in a very large network how gnutella2 would be any different than gnutella today.


Because Gnutella today limits searches with a TTL (I believe a default of 7), so there is no possible way to search the whole network even if you let it search for years. G2 is not limited by that TTL. Sure searching 1 million users is possible. Of course it will take longer to search 1 milion than 500,000, but that doesnt make it any less possible. There is no true P2P app out there than can claim the same thing.

[Anenga2]

Originally posted by Sephiroth

Also i think the last thing you should do is to speak for other developers who have not said that they will support it, it would be another stupid move by you to start telling people "oh theyll support it if its stable" and like i said it would be incredibly stupid for them to say that they will support anything without even seeing it..
I wrote that based on the reactions in the GDF. Vinnie wrote on the BearShare forum he'll support it if Limewire does. And the rest I got from the GDF replies.

They will see it eventually.


[Mikes post your referring too has a few problems. First limewire's host count is undercounting the network by about at least 4 times. So the real number of hosts on gnutella is 400000 users and its most likely much higher than that not 100000. So if gnutella moved to "gnutella2" then a person would not be able to search the entire network. It may support global searching on a 500,000 user network but what about millions? After 500,000 users it seems that "gnutella2" suffers many problems that the current network. How can gnutella compete with a network like fasttrack if they can support millions of users yet "g2" only guarentees up to support only 500000 and the current gnutella size is around that? So is it really that much better as you are marketing as it being? Keep in mind im baseing all this off the information in mikes post..
Why must you be so negative?

Mike said it could handle 500,000. He didn't say that's all it could handle. Considering the fact that G2 Hubs can handle easily 500 Leafs as the equilavant to 250 that G1 Ultrapeers can handle, I'd say that the G2 Network could handle a userbase up in the millions.

[Anenga2]

Originally posted by Sephiroth
Its not really known for sure since fasttrack limits search results. Even though fasttrack was probably a poor example I still would like to know given in a very large network how gnutella2 would be any different than gnutella today.
Well tell everyone and your mother to get Shareaza and eventually maybe we can find out the answer for you.


In any case one of the main parts of "gnutella2" is this marketing.. Which Gnutella does not need to be "rebranded" repackaged, sugarcoated and forced down people's throats, people dont like gnutella because they think its centralized and dont know anything about it. The majority of people who run around and say "gnutella sucks" have either never used a gnutella program ever or have not in a very very long time.
"Main Parts"? It is a theme of Gnutella2, and is a reason why Mike called it "Gnutella2" but that isn't the only reason for Gnutella2, or what Gnutella2 is all about. Gnutella needed a new refreshed image. A few weeks ago everyone hated it, now it seems that people have hope in it. The average person doesn't know what the hell Gnutella is anyways.

I could go on and on on the benifits of Gnutella2 vs. Gnutella1 or FastTrack or eDonkey or whatever. But it will all go over your head. Just wait until the specs are out before you spread FUD about G2 when you really do not know anything about it. The positive descriptions Mike, I and others have given out are easily identifiable if you use Gnutella2 in Shareaza. There have yet to be no legitimate defense against Mike's claims.

[Sephiroth]

Originally posted by Anenga2

Why must you be so negative?

Mike said it could handle 500,000. He didn't say that's all it could handle. Considering the fact that G2 Hubs can handle easily 500 Leafs as the equilavant to 250 that G1 Ultrapeers can handle, I'd say that the G2 Network could handle a userbase up in the millions.

Im not negative im skeptical of it. No offense but this isnt the first time that someone has come along saying that they are going to revolutionize p2p and didnt.. The first "gnutella2" is a classic example and there are numerous others too..

I didnt say that it couldnt handle more but would it be as efficient. Gnutella now with enough ultrapeers could support millions but search wise it wouldnt be that efficient and that is the question im asking about gnutella2. With a large network in the millions can it still perform the same as you are marketing it as? Sure it may be a improvement to gnutella but its the big picture that matters. There were over 20 million people using napster at one time.

rainwater I allready know how the searches are done on the current network broadcasted out with a TTL and etc.. Which of course it wont search the entire network because the message dies after 7 hops and its Time To Live expires. Reason why programs end searches is that they could go on forever and that would use up a tremendous amount of network resources and would pretty likly kill the network because the bandwidth that it demands would just be forever increaseing.

[rainwater]

Originally posted by Sephiroth

There were over 20 million people using napster at one time.


You do realize that Napster was not Peer to Peer when it came to searching? And as I recall, you could only search users that were connected to the same server as you. I really don't see you point. G2 is able to search the whole network, plain and simple.

[Sephiroth]

Originally posted by rainwater


You do realize that Napster was not Peer to Peer when it came to searching? And as I recall, you could only search users that were connected to the same server as you. I really don't see you point. G2 is able to search the whole network, plain and simple.

Yes i know i used napster since the first verison and only had 2 servers with about 5000 people total users. They linked half of servers and the other half so there were two part you can search which totalsaround75-80 servers.

The point i was trying to make is that thos 20 million people who have used P2P before for a while are out there somewhere. So its not unrealistic for a p2p program to have millions of people using it which again was my point not the technical aspects of napster..

No its not as pain and simple as you think it is..

[Anenga2]

Originally posted by Sephiroth


Yes i know i used napster since the first verison and only had 2 servers with about 5000 people total users. They linked half of servers and the other half so there were two part you can search which totalsaround75-80 servers.
Okay, you don't have to prove yourself to us.


The point i was trying to make is that thos 20 million people who have used P2P before for a while are out there somewhere. So its not unrealistic for a p2p program to have millions of people using it which again was my point not the technical aspects of napster..
So what your saying is that if Shareaza/Gnutella2 cannot handle 20+ million users on the network and return lightning fast results, then it's a piece of crap and we should keep with the current Gnutella?

Shareaza, Mike, nobody said Shareaza could handle 20 million nodes. Though, I bet it could scale a hell of a lot better than G1 or any other P2P Network out there with 20 million nodes, but Mike would have to look into it when that time comes. And it doesn't seem to be anytime soon.

[Crazy Horse]

Damn!!! Three days - Three updates. These guys are pretty busy. 1.7.0.8 out today.

[Fatboyslack54]

YES YOU ARE NEGATIVE.......


You have to understand, every damn reply you make has something that pisses someone off. No offence or anything, i like most of your posts, but sometimes like on this one you know how to get to a person real bad. You did act negative as Annenga tried and Rainwater tried to explain something to you BUT NO i have a different idea or version of what you are saying and MINE IS RIGHT!! Dude i think they know more about gnutella and Shareaza than you do so just stick with what they say!!:devil

[crackerjacker]

if u can go on a quest on a google search engine u can fine some good infomration on this topic.
research is important on anyting.
like with any new version of gnutella programs, i think it is if the program works for someone its cool.
i tried lots of gnutella clients, except macintosh versions, and linux.
it seems shareaza is getting awesome reviews becuase of the updates. now i like shareaza, feature enrich. and people who test the program, and like it, use it. it seems
but i still would love more information on the g2
heh, at least people still work on gnutella, al tho people dont have to agree with all the stuff of gnutella or other p2p programs.
isnt the real issue that it gets the job done*

rtws

[FutureGenetics]

Originally posted by evilmegaman
One thing is that gnutella2 is like shareaza's way of doing what bearshare does. I think it is just a portion of the gnutella network as I still get uploads from clients like gnucleus 1.6.0.0 and others so there is something very sneaky about gnutella2. I think that mike isn't like he used to be at all... He's trying to hog the glory... he has become a bit overwhelmed... It is very sad for me to say this but... I think he is lying about gnutella2.... I think it is shareaza's portion of gnutella like bearshare did. you can still upload to other clients but you can only connect to a small number of other things...

in reply to nothingface, He is a little skeptical (http://www.gnutellaforums.com/showthread.php?s=&threadid=17017)


You can turn off connecting to the G1 protocol and just connect to nodes that use the G2 protocol. By default shareaza connects to both, whatever it can snag in the connection process.